Multiscale Region-Level VHR Image Change Detection via Sparse Change Descriptor and Robust Discriminative Dictionary Learning

Table 1

Performance comparison of different CD methods. The bold value is the best and the italic one is the second.

Data set	Method	FAR (%)	MAR (%)	TFR (%)	Method	FAR (%)	MAR (%)	TFR (%)
DS1	pxmsCRF	4.04	12.96	6.60	pxlinSVM	19.24	85.32	38.20
	pxrbfSVM	6.39	18.33	9.81	splinSVM	9.72	26.59	14.56
	sprbfSVM	4.76	16.91	8.24	pKNN	2.10	22.99	8.09
	pKNN-CSR	2.54	17.50	6.83	pKNN-CRF	0.61	18.56	5.76
DS2	pxmsCRF	7.13	18.85	8.89	pxlinSVM	9.42	41.23	14.19
	pxrbfSVM	14.18	19.49	14.97	splinSVM	3.32	48.51	10.11
	sprbfSVM	5.67	31.14	9.49	pKNN	6.87	29.63	10.29
	pKNN-CSR	5.74	25.34	8.68	pKNN-CRF	1.84	24.48	5.24
DS3	pxmsCRF	20.76	7.20	18.95	pxlinSVM	9.96	65.42	17.36
	pxrbfSVM	12.07	34.73	15.09	splinSVM	3.63	59.77	11.13
	sprbfSVM	0.40	83.81	11.53	pKNN	13.93	11.85	13.65
	pKNN-CSR	5.77	19.04	7.55	pKNN-CRF	3.25	11.03	4.29